A high-definition blood vessel structure reconstruction and blood flow function evaluation method and system

Abstract

The application provides a high-definition blood vessel structure reconstruction and blood flow function evaluation method and system, constructs a high-dimensional feature manifold space based on an ultrasonic contrast image, identifies a blood vessel subgroup with different blood flow function modes, finally obtains a high-definition rainbow blood vessel structure video and a blood vessel heterogeneity index in an ultrasonic scanning area, can not only provide a powerful tool for evaluating tumor microvessel structure, blood flow function and spatiotemporal heterogeneity characteristics, provide rich parameters for quantitatively describing a complex blood vessel system of a tumor for early evaluation of drug efficacy, and can specifically identify a target blood perfusion area according to a function characteristic, provide a target prompt and feedback intervention basis for drug targeted release.

Description

Technical Field

[0001] This application relates to the field of ultrasound imaging technology, specifically to a high-definition vascular structure reconstruction and blood flow function evaluation method, a high-definition vascular structure reconstruction and blood flow function evaluation device, as well as electronic equipment and computer-readable storage media. Background Technology

[0002] Angiogenesis is one of the key biomarkers of tumors. Since tumors rely on the formation of microvascular systems for growth and metastasis, dynamic and clear observation of microvessels is of great significance for clinical diagnosis and treatment evaluation. A growing body of clinical studies demonstrates that almost all malignant tumors exhibit significant spatiotemporal heterogeneity on imaging, and greater heterogeneity often reflects treatment resistance in drug-resistant cell subpopulations, indicating negative drug treatment effects. In fact, changes in the microenvironment caused by abnormal microblood flow, such as hypoxia and acidosis, can stimulate the production of more metastatic and drug-resistant cell subpopulations in local tumor areas, which can significantly weaken the effectiveness of traditional global cancer treatment strategies.

[0003] In recent years, targeted precision therapy strategies capable of controlled release under endogenous environmental characteristics (hypoxia and / or enzyme activity) or external triggers (light / acoustic / magnetic) have gradually attracted attention. These strategies aim to improve efficacy and reduce toxicity by maximizing the accumulation of free drugs at predetermined sites of action. However, determining the specific location of drug release in spatiotemporally dynamic lesions (e.g., tumors) remains a significant challenge for targeted therapy. Therefore, developing spatially explicit imaging modalities and guiding tumor-targeted therapy strategies holds promise for achieving a synergistic effect greater than the sum of its parts (1+1>2), improving the feasibility of tumor cure and preventing metastasis and recurrence, thus presenting tremendous opportunities for the development of new tumor treatment methods. However, traditional magnetic resonance imaging, computed tomography, or Doppler ultrasound imaging have insufficient spatial resolution, making it difficult to clearly image actively changing tumor microvessels and specifically identify highly malignant tumor regions.

[0004] In recent years, ultrasound imaging technology has developed rapidly. In particular, contrast-enhanced ultrasound imaging, using microbubble contrast agents, has achieved high signal-to-background ratio enhancement of minute blood flow signals. Furthermore, ultrasound super-resolution imaging based on contrast-enhanced ultrasound imaging can non-invasively reconstruct the morphology and blood flow function of tumor microvessels. To date, however, this technology has not been directly applied to target area identification for drug release. Summary of the Invention

[0005] This application provides a high-definition vascular structure reconstruction and blood flow function evaluation method and system, a high-definition vascular structure reconstruction and blood flow function evaluation device, as well as electronic equipment and computer-readable storage medium. It can not only reliably evaluate the tumor microvascular structure, blood flow function and spatiotemporal heterogeneity characteristics, providing rich parameters for quantitative characterization of the complex tumor vascular system for early evaluation of drug efficacy, but also identify the target blood perfusion area according to functional characteristics, providing target prompts and feedback intervention basis for targeted drug release.

[0006] This application provides a high-definition vascular structure reconstruction and blood flow function evaluation method. It constructs a high-dimensional feature manifold space based on ultrasound contrast imaging images, automatically identifies vascular subgroups with different blood flow function patterns, and ultimately obtains a high-definition rainbow vascular structure video and vascular heterogeneity indicators within the ultrasound scanning area. Specific steps include: stitching the original video Z0 containing M frames of ultrasound contrast imaging images sequentially to obtain a video space Ω1; further, identifying the pixels to which microbubbles belong to obtain a total of R microbubble spatiotemporal trajectory pixels; within the video space Ω1, estimating the instantaneous velocity and direction of microbubble motion at each microbubble spatiotemporal trajectory pixel, i.e., instantaneous blood flow function characteristics; and then, based on the R... A high-dimensional feature manifold space Ω2 is constructed by using the instantaneous blood flow functional features corresponding to the microbubble spatiotemporal trajectory pixels. Within the high-dimensional feature manifold space Ω2, the instantaneous blood flow functional features corresponding to the R microbubble spatiotemporal trajectory pixels are classified into K classes, i.e., K types of blood flow functional modes, using a pattern recognition method. The blood flow functional modes to which the instantaneous blood flow functional features corresponding to each microbubble spatiotemporal trajectory pixel belong are unique. The instantaneous blood flow functional features corresponding to the R microbubble spatiotemporal trajectory pixels classified into K classes in the high-dimensional feature manifold space Ω2 are backtracked to the R microbubble spatiotemporal trajectory pixels in the video space Ω1, and each microbubble spatiotemporal trajectory pixel is assigned a unique color code C according to its category. i (i = 1, 2, 3, ..., K); subsequently, K rainbow ultrasound contrast imaging videos Z0 of the same size as the original video Z0 are formed. i (i = 1, 2, 3, ..., K); for the rainbow ultrasound contrast video Z i Super-resolution reconstruction was performed to obtain a high-definition video of the vascular structure. * The number of categories K and the inter-class differences were used as indicators of vascular heterogeneity within the scanning field of view.

[0007] In one embodiment, the original video Z0 containing M frames of ultrasound contrast images is stitched together in chronological order to obtain the video space Ω1. Further, the process of identifying the pixels to which microbubbles belong to obtain a total of R microbubble spatiotemporal trajectory pixels includes: stitching the original video Z0 (image size A×L, unit: pixels) containing M frames of two-dimensional ultrasound contrast images frame by frame in chronological order as a numerical matrix to construct the video space Ω1 (three-dimensional, A×L×M, unit: pixels); and identifying the R microbubble spatiotemporal trajectory pixels P(x) within the video space Ω1 whose grayscale values ​​are greater than a set grayscale threshold. r , z r , t r )(r=1,2,3,...,R,1≤x r ≤A, 1≤y r ≤L,1≤z r ≤M) are identified as pixels belonging to the spatiotemporal trajectory of microbubbles, wherein the R microbubble spatiotemporal trajectory pixels P(x r , z r , t r It can present the three-dimensional spatiotemporal trajectory of moving microbubbles in the video space Ω1.

[0008] In one embodiment, estimating the instantaneous velocity and direction of microbubbles at each microbubble spatiotemporal trajectory pixel point within the video space Ω1, i.e., the instantaneous blood flow functional characteristics, includes: in the video space Ω1, based on R microbubble spatiotemporal trajectory pixels P(x r , z r , t r The distribution gradient of local gray values ​​at (r = 1, 2, 3, ..., R) is calculated, and the local structure tensor is calculated; and / or the trajectory skeleton is extracted and a straight line is fitted to obtain R microbubble spatiotemporal trajectory pixels P(x) r , z r , t r The instantaneous trajectory direction vector T at (r = 1, 2, 3, ..., R) r (x r , z r , t r (r = 1, 2, 3, ..., R); Calculate the trajectory direction vector T for each instantaneous moment. r (x r , z r , t r The azimuth angle θ corresponding to (r = 1, 2, 3, ..., R) r (x r , z r , t r (r = 1, 2, 3, ..., R) and polar angle The spatiotemporal trajectories of R microbubbles in the video space Ω1 are obtained as follows: P(x) r , z r , t r The instantaneous motion direction D of the microbubbles at (r = 1, 2, 3, ..., R) r (r = 1, 2, 3, ..., R) and instantaneous velocity V r (r = 1, 2, 3, ..., R);

[0009] In one embodiment, a high-dimensional feature manifold space Ω2 is constructed based on the instantaneous blood flow functional features corresponding to R microbubble spatiotemporal trajectory pixels, including: using the instantaneous motion direction and instantaneous motion rate as a basis, the R microbubble spatiotemporal trajectory pixels P(x) in the video space Ω1 are... r , z r , t r The instantaneous motion direction D of the microbubbles at (r = 1, 2, 3, ..., R) r (r = 1, 2, 3, ..., R) and instantaneous velocity V r (r = 1, 2, 3, ..., R) are paired to form the instantaneous blood flow functional feature vector {(D r V r )|r=1,2,3,...,R} and mapped to the high-dimensional feature manifold space Ω2.

[0010] In one embodiment, within the high-dimensional feature manifold space Ω2, a pattern recognition method is used to classify the instantaneous blood flow functional features corresponding to R microbubble spatiotemporal trajectory pixels into K classes, including: the instantaneous blood flow functional feature vectors {(D r V r Unsupervised or weakly supervised clustering is performed on the sequence |r=1,2,3,...,R} to obtain K clusters, which are K types of blood flow function patterns. Each instantaneous blood flow function feature vector belongs to a unique blood flow function pattern, that is, each microbubble spatiotemporal trajectory pixel belongs to a unique blood flow function pattern.

[0011] In one embodiment, the instantaneous blood flow functional features corresponding to the R microbubble spatiotemporal trajectory pixels in the high-dimensional feature manifold space Ω2, which are divided into K classes, are backtracked to the R microbubble spatiotemporal trajectory pixels in the video space Ω1, and each microbubble spatiotemporal trajectory pixel is assigned a unique color code C according to its class. i (i = 1, 2, 3, ..., K); subsequently, K rainbow ultrasound contrast imaging videos Z0 of the same size as the original video Z0 are formed. i (i = 1, 2, 3, ..., K) includes: tracing back R microbubble spatiotemporal trajectory pixels from the high-dimensional feature manifold space Ω2 to the video space Ω1 one by one, and based on the R microbubble spatiotemporal trajectory pixels P(x) in the high-dimensional feature manifold space Ω2...r , z r , t r The instantaneous blood flow functional feature vector {(D)} r V r The unique blood flow function mode of the sequence Ω1 (r = 1, 2, 3, ..., R) assigns a unique color code C to the corresponding microbubble spatiotemporal trajectory pixel in the video space Ω1. i (i = 1, 2, 3, ..., K), ultimately forming K subsets of microbubble spatiotemporal trajectory pixels; according to the color encoding order, the pixels in the original video Z0 belonging to color encoding C... i The subset of pixels representing the spatiotemporal trajectory of the i-th microbubble (i = 1, 2, 3, ..., K) is sequentially separated to construct K rainbow ultrasound contrast imaging videos Z0 of the same size as the original video Z0. i (i = 1, 2, 3, ..., K).

[0012] In one embodiment, the rainbow ultrasound contrast video Z i Super-resolution reconstruction was performed to obtain a high-definition video of the vascular structure. * Including, Z of K rainbow ultrasound contrast images i (i = 1, 2, 3, ..., K) are reconstructed using ultrasound localization microscopy, resulting in K color codes coded as C. i Super-resolution microvessel subset image V (i = 1, 2, 3, ..., K) i (i = 1, 2, 3, ..., K), further, by analyzing all super-resolution microvascular subset images V i Weighted superposition yields a rainbow-like super-resolution microvascular image with K colors, i.e., a high-definition vascular structure video Z. * .

[0013] In one embodiment, the number of categories K and the inter-class differences are used as indicators of vascular heterogeneity within the scanning field of view. This includes calculating the maximum / minimum distance between the centers of the K clusters based on the number of blood flow function patterns K in the high-dimensional feature manifold space Ω2 and the spatial distribution of the K blood flow function patterns; and / or the uniformity of the proportion of the K clusters in the total.

[0014] This application provides a high-definition vascular structure reconstruction and blood flow function evaluation system, which consists of the following modules:

[0015] The video space construction module stitches together the original video Z0 containing M frames of ultrasound contrast images in chronological order to obtain the video space Ω1. Furthermore, by identifying the pixels to which the microbubbles belong, a total of R microbubble spatiotemporal trajectory pixels are obtained.

[0016] The high-dimensional feature manifold space construction module estimates the instantaneous motion rate and direction of each microbubble at the spatiotemporal trajectory pixel point in the video space Ω1, i.e. the instantaneous blood flow function features. Then, based on the instantaneous blood flow function features corresponding to R microbubble spatiotemporal trajectory pixels, the high-dimensional feature manifold space Ω2 is constructed.

[0017] The blood flow function pattern recognition module uses a pattern recognition method to classify the instantaneous blood flow function features corresponding to R microbubble spatiotemporal trajectory pixels into K classes, i.e. K types of blood flow function patterns, within the high-dimensional feature manifold space Ω2. Among them, the blood flow function feature corresponding to each microbubble spatiotemporal trajectory pixel belongs to a unique blood flow function pattern.

[0018] The Rainbow ultrasound contrast imaging video reconstruction module backtracks the instantaneous blood flow functional features corresponding to the R microbubble spatiotemporal trajectory pixels, which are divided into K classes in the high-dimensional feature manifold space Ω2, to the R microbubble spatiotemporal trajectory pixels in the video space Ω1, and assigns a unique color code C to each microbubble spatiotemporal trajectory pixel according to its class. i (i = 1, 2, 3, ..., K); subsequently, K rainbow ultrasound contrast imaging videos Z0 of the same size as the original video Z0 are formed. i (i = 1, 2, 3, ..., K);

[0019] High-definition vascular structure video reconstruction module: This module reconstructs rainbow ultrasound contrast images. i High-resolution vascular structure video Z was obtained using a super-resolution reconstruction algorithm. * ;

[0020] Vascular heterogeneity assessment module: The number of classes K within the scanning field of view and the inter-class differences are used as indicators of vascular heterogeneity.

[0021] Accordingly, this application provides an electronic device, including: a processor; and a memory storing computer program instructions, which, when executed by the processor, cause the processor to perform the high-definition vascular structure reconstruction and blood flow function evaluation method and system as described above.

[0022] Accordingly, this application provides a computer-readable storage medium storing computer program instructions, which, when executed by a processor, cause the processor to perform the high-definition vascular structure reconstruction and blood flow function evaluation method and system as described above.

[0023] Benefits

[0024] This application provides a high-definition vascular structure reconstruction and blood flow function evaluation method, a high-definition vascular structure reconstruction and blood flow function evaluation device, as well as an electronic device and a computer-readable storage medium. Based on the high-dimensional feature manifold space constructed from ultrasound contrast imaging video, various blood flow function patterns and high-definition rainbow vascular structure videos are identified. Simultaneously, the tumor microvascular structure, blood flow function, and spatiotemporal heterogeneity characteristics are evaluated. This not only provides a powerful tool for early clinical treatment evaluation but also provides target indications and feedback intervention basis for targeted drug release. Attached Figure Description

[0025] Figure 1 The diagram shown is a flowchart of a high-definition vascular structure reconstruction and blood flow function evaluation method provided in an embodiment of this application.

[0026] Figure 2A The image shown is a schematic diagram of a video space including multiple microbubbles according to an embodiment of this application.

[0027] Figure 2B The image shown is a video spatial diagram including multiple microbubble spatiotemporal trajectories provided in an embodiment of this application.

[0028] Figure 3 The diagram shown is a schematic diagram of the calculation of the instantaneous trajectory direction vector and instantaneous blood flow functional characteristics at a single microbubble spatiotemporal trajectory pixel point provided in an embodiment of this application.

[0029] Figure 4 The diagram shown is a schematic diagram of the instantaneous blood flow function feature vector of a single microbubble spatiotemporal trajectory pixel in a high-dimensional feature manifold space provided by an embodiment of this application.

[0030] Figure 5A The diagram shown is a schematic diagram of the instantaneous blood flow function feature vector of 48,642 microbubble spatiotemporal trajectory pixels in a high-dimensional feature manifold space provided in an embodiment of this application.

[0031] Figure 5B The diagram shown is a schematic representation of the blood flow function pattern recognition result of the instantaneous blood flow function feature vector of 48,642 microbubble spatiotemporal trajectory pixels in a high-dimensional feature manifold space provided by an embodiment of this application.

[0032] Figure 6 The diagram shown is a schematic representation of the blood flow function pattern recognition results of 48,642 microbubble spatiotemporal trajectory pixels in the video space provided in an embodiment of this application.

[0033] Figure 7A The diagram shown is a schematic diagram of the spatiotemporal trajectory of microbubbles corresponding to the first blood flow function mode in the video space provided in an embodiment of this application.

[0034] Figure 7BThe diagram shown is a schematic diagram of the spatiotemporal trajectory of microbubbles corresponding to the second blood flow functional mode in the video space provided in an embodiment of this application.

[0035] Figure 7C The diagram shown is a schematic diagram of the spatiotemporal trajectory of microbubbles corresponding to the third blood flow functional mode in the video space provided in an embodiment of this application.

[0036] Figure 7D The diagram shown is a schematic diagram of the spatiotemporal trajectory of microbubbles corresponding to the fourth blood flow functional mode in the video space provided in an embodiment of this application.

[0037] Figure 8A The image shown is a schematic diagram of a super-resolution microvascular subset image corresponding to the first blood flow function mode in the video space provided in an embodiment of this application.

[0038] Figure 8B The image shown is a schematic diagram of a super-resolution microvascular subset image corresponding to the second blood flow functional mode in the video space provided in an embodiment of this application.

[0039] Figure 8C The image shown is a schematic diagram of a super-resolution microvascular subset image corresponding to the third blood flow functional mode in the video space provided in an embodiment of this application.

[0040] Figure 8D The image shown is a schematic diagram of a super-resolution microvascular subset image corresponding to the fourth blood flow functional mode in the video space provided in an embodiment of this application.

[0041] Figure 9 The diagram shown is a schematic diagram of a method for identifying the target area of ​​drug action and focusing drug release for high-definition vascular structure reconstruction and blood flow function evaluation provided in an embodiment of this application.

[0042] Figure 10 The diagram shown is a schematic diagram of the high-definition vascular structure reconstruction and blood flow function evaluation device provided in an embodiment of this application.

[0043] Figure 11 The diagram shown is a structural schematic of an electronic device provided in an embodiment of this application. Detailed Implementation

[0044] The technical solutions in the embodiments of this application will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0045] As mentioned earlier, changes in the tumor microenvironment caused by abnormal microblood flow, such as hypoxia and acidosis, can stimulate the production of more metastatic and drug-resistant cell subpopulations in local tumor areas. This can significantly weaken the effectiveness of traditional global cancer treatment strategies. Although controlled-release targeted precision therapy strategies can improve efficacy and reduce toxicity by maximizing the accumulation of free drugs at predetermined sites of action, determining the specific location of drug release in spatiotemporally dynamic tumors remains a major challenge for targeted therapy. Therefore, developing spatially explicit imaging modalities and their guided tumor-targeted therapy strategies is crucial for clinical precision medicine. However, traditional clinical imaging strategies not only fail to clearly image complex tumor microvessels but also cannot specifically identify tumor subregions with high malignant potential.

[0046] To address the aforementioned technical issues, this application proposes a high-definition vascular structure reconstruction and blood flow function evaluation method and system. The basic idea is as follows: only multiple frames of ultrasound contrast imaging containing flowing microbubble images need to be captured. Then, a high-dimensional feature manifold space is constructed based on the multiple frames of ultrasound contrast imaging including microbubble images. This allows for the identification of vascular subgroups with different blood flow function patterns, ultimately obtaining a high-definition rainbow vascular structure video and vascular heterogeneity indicators within the ultrasound scanning area. This enables reliable assessment of tumor microvascular structure, blood flow function, and spatiotemporal heterogeneity characteristics, providing rich parameters for quantitative characterization of the complex vascular system of tumors for early evaluation of drug efficacy. Furthermore, it can specifically identify target blood perfusion subregions according to functional characteristics, thereby providing target indications and feedback intervention basis for targeted drug release.

[0047] It should be noted that this application does not specifically limit the application scenarios for high-definition vascular structure reconstruction and blood flow function evaluation. After introducing the basic principles of this application, various non-limiting embodiments of this application will be described in detail below.

[0048] Example 1

[0049] Figure 1 The diagram shown is a flowchart illustrating a high-definition vascular structure reconstruction and blood flow function evaluation method according to an embodiment of this application. Figure 1 As shown, this high-definition vascular structure reconstruction and blood flow function evaluation method includes the following steps:

[0050] Step 101: The original video Z0 containing M frames of ultrasound contrast images is spliced ​​together in chronological order to obtain the video space Ω1. Furthermore, a total of R microbubble spatiotemporal trajectory pixels are obtained by identifying the pixels to which the microbubbles belong.

[0051] Specifically, the original video Z0 containing M frames of ultrasound contrast images can be an image captured by an ultrasound imaging system after an ultrasound contrast agent is injected into the blood. The image of the microbubble in the ultrasound contrast image can be a point spread function image spreading out with a certain radius around the microbubble. Due to the diffraction effect of the ultrasound waves emitted by the ultrasound imaging system, the microbubble appears in the contrast image as an image spreading out with a certain radius around the microbubble, that is, an image of the microbubble's point spread function.

[0052] In each frame of ultrasound contrast imaging, each microbubble has a corresponding point spread function image. When the microbubble flows in the blood vessel, after stitching together the two-dimensional ultrasound contrast imaging images of size A×L in M ​​frames, which include multiple microbubble images, in the form of a numerical matrix and in chronological order, a three-dimensional video space Ω1 of size A×L×M can be formed, which includes information in two spatial dimensions and one temporal dimension. Since the microbubble images of the flowing microbubbles can be partially connected and overlapped in the temporal dimension (dimension size M) of the video space Ω1, each microbubble can form a coherent three-dimensional columnar structure in the video space Ω1.

[0053] Figure 2A The diagram shown is a schematic representation of how the original video Z0 is stitched together in chronological order to obtain the video space Ω1, according to an embodiment of this application. Figure 2A As shown, video space 220 is obtained by stitching together the original video containing 300 frames of ultrasound contrast images of size 100×100 pixels in chronological order. The size is 100×100×300. It includes 8 three-dimensional columnar structures formed by 8 microbubbles, namely columnar structure 211, columnar structure 212, columnar structure 213, columnar structure 214, columnar structure 215, columnar structure 216, columnar structure 217, and columnar structure 218.

[0054] It should be understood that the number of microbubbles in each frame of ultrasound contrast imaging can be selected according to actual needs, such as reconstruction accuracy, blood flow velocity in the imaging area and vascular complexity. This application does not make a specific limit on the number of microbubbles in each frame of ultrasound contrast imaging.

[0055] In one embodiment, step 101 can also involve stitching together 200, 300, or 400 consecutive frames of ultrasound contrast images within a time interval in chronological order. Specifically, the number of frames stitched in chronological order can be selected based on actual conditions. For example, when the blood flow velocity in the imaging area is slow, the acquisition time can be extended and the number of frames acquired can be increased. Simultaneously, the ultrasound imaging frame rate can be selected based on actual conditions. For example, when the blood flow velocity in the imaging area is fast, i.e., the microbubble movement velocity is fast, the ultrasound imaging frame rate can be increased, thereby controlling the microbubbles to form a three-dimensional connected columnar structure in the video space.

[0056] It should be understood that, due to the interference of noise and background signals in ultrasound contrast images, it is necessary to distinguish pixels belonging to microbubbles from noise and background pixels in the video space. Specifically, this can be achieved by... Figure 2A The columnar structures 211 to 218 shown are used for microbubble spatiotemporal trajectory pixel identification to obtain... Figure 2B The microbubble spacetime trajectories shown are 221, 222, 223, 224, 225, 226, 227, and 228.

[0057] In one embodiment, 48,642 microbubble spatiotemporal trajectory pixels P(x) within the video space Ω1 with gray values ​​greater than a set gray value threshold of 0.1 (gray value range 0-1) are selected. r , z r , t r )(r=1, 2, 3,..., 48642, 1≤x r ≤100, 1≤y r ≤100, 1≤z r Pixels with a value ≤300 were identified as belonging to the spatiotemporal trajectory of microbubbles, while the remaining 2,951,358 pixels were identified as non-microbubble spatiotemporal trajectory pixels. Figure 2B As shown, the 48,642 microbubble spatiotemporal trajectory pixels P(x) r , z r , t r It can present the three-dimensional spatiotemporal trajectory of moving microbubbles in the video space Ω1.

[0058] It should be understood that the grayscale threshold can be set according to actual needs, such as the signal-to-noise ratio of ultrasound contrast images, the degree of respiratory motion interference, etc. This application does not make specific limitations on setting the grayscale threshold.

[0059] Step 102: Within the video space Ω1, estimate the instantaneous motion speed and direction of each microbubble at the spatiotemporal trajectory pixel point, i.e. the instantaneous blood flow function features. Then, construct a high-dimensional feature manifold space Ω2 based on the instantaneous blood flow function features corresponding to R microbubble spatiotemporal trajectory pixels.

[0060] Specifically, since the spatiotemporal trajectory of a microbubble is obtained by stitching together multiple frames of contrast images within a time interval in chronological order, the spatiotemporal trajectory of a microbubble can be considered as the motion trajectory of the microbubble within the time interval containing the multiple frames of contrast images. Therefore, the direction and velocity of a microbubble at any position on its spatiotemporal trajectory are its instantaneous direction and velocity, and the instantaneous direction and velocity of the microbubble corresponding to each pixel coordinate on its spatiotemporal trajectory are equal to the instantaneous direction and velocity of the corresponding microbubble in the corresponding time point and frame.

[0061] For example, it can be in the case of Figure 2B The 48,642 microbubble spatiotemporal trajectory pixels P(x) on microbubble spatiotemporal trajectories 221 to 228 shown are illustrated. r , z r , t r )(r=1, 2, 3,..., 48642, 1≤x r ≤100, 1≤y r ≤100, 1≤z r Calculate the structure tensor at ≤300; and / or extract the trajectory skeleton and perform linear fitting to obtain 48642 microbubble spatiotemporal trajectory pixels P(x r , z r , t r The instantaneous trajectory direction vector T at (r = 1, 2, 3, ..., 48642) r (x r , z r , t r )(r=1, 2, 3,..., 48642);

[0062] In one embodiment, the instantaneous trajectory direction vector T is obtained by extracting the trajectory skeleton and fitting a straight line. r Specifically, the trajectory skeleton can be extracted using an iterative erosion boundary algorithm. This algorithm obtains the skeleton of the corresponding microbubble spatiotemporal trajectory by performing multiple refinement operations. The algorithm is simple and computationally fast. Line fitting can be performed using the Hough transform detection method. This method has strong noise and deformation resistance and can accurately extract the trajectory line corresponding to the trajectory.

[0063] In one embodiment, the structure tensor can be obtained using the dominant gradient direction in the neighborhood of a pixel. Based on the dominant local gray-level gradient direction indicated by the structure tensor, the instantaneous trajectory direction vector T perpendicular to it is calculated. r .

[0064] Furthermore, the spatiotemporal trajectory pixels P(x) of 48642 microbubbles in the video space Ω1 are calculated. r , z r , t r (The instantaneous motion directions D of the 48642 microbubbles at r = 1, 2, 3, ..., 48642) r (r = 1, 2, 3, ..., 48642) and 48642 instantaneous motion rates V r (r = 1, 2, 3, ..., 48642), when paired, form 48642 instantaneous blood flow functional feature vectors {(D r V r)|r=1,2,3,...,48642}, and map the 48642 instantaneous blood flow function feature vectors to the high-dimensional feature manifold space Ω2.

[0065] Specifically, the spherical coordinates of the instantaneous trajectory direction vector at the pixel point of the microbubble spatiotemporal trajectory can be calculated first, and then the instantaneous motion direction and instantaneous motion velocity of the microbubble at the pixel point of the microbubble spatiotemporal trajectory can be calculated, ultimately forming an instantaneous blood flow functional feature vector. For example... Figure 3 As shown, the instantaneous trajectory direction vector 31 at a single microbubble spatiotemporal trajectory pixel 30 is represented in spherical coordinates, yielding the polar angle 33 and azimuth angle 32. In one embodiment, since the instantaneous axial motion direction of the microbubble at the spatiotemporal trajectory pixel 30 is along the positive y-axis, the instantaneous axial motion direction of the microbubble is set to -1; otherwise, it is set to 1. The γ value is used as the instantaneous lateral motion direction of the microbubble, and the instantaneous axial motion direction and the instantaneous lateral motion direction are combined to form the instantaneous motion direction of the microbubble, i.e., D. 30 = [-1, γ]; the absolute value of 1 / tan(θ) is taken as the instantaneous velocity of the microbubble at pixel 30 of the microbubble spatiotemporal trajectory, i.e., V r =|1 / tan(θ)|. In summary, the instantaneous blood flow function feature vector [-1, γ, |1 / tan(θ)|] corresponding to the microbubble spatiotemporal trajectory pixel 30 is obtained.

[0066] Repeating the above steps yields a total of 48,642 microbubble spatiotemporal trajectory pixels P(x). r , z r , t r 48642 instantaneous blood flow functional feature vectors {(D) (r = 1, 2, 3, ..., 48642) r V r )|r=1, 2, 3,..., 48642}.

[0067] by Figure 3 Taking the microbubble spatiotemporal trajectory pixel 30 as an example, as shown Figure 4 As shown, in the high-dimensional feature manifold space 40 composed of the microbubble instantaneous velocity axis 41, the microbubble instantaneous axial motion direction axis 42, and the microbubble instantaneous transverse motion direction axis 43, Figure 3 The microbubble spatiotemporal trajectory pixel 30 corresponds to the position projected onto feature point 44, with corresponding coordinates in the high-dimensional feature manifold space Ω2 as [-1, γ, |1 / tan(θ)|]. Repeating the above steps, a total of 48642 microbubble spatiotemporal trajectory pixels P(x) are obtained in the high-dimensional feature manifold space 40. r , z r , t r 48642 instantaneous blood flow functional feature points (r = 1, 2, 3, ..., 48642), such as Figure 5AAs shown, the coordinates correspond to {(D)} r V r )|r=1, 2, 3,..., 48642}.

[0068] Step 103: Within the high-dimensional feature manifold space Ω2, the instantaneous blood flow functional features corresponding to the R microbubble spatiotemporal trajectory pixels are classified into K classes, i.e., K types of blood flow functional modes, using a pattern recognition method.

[0069] Specifically, steps 101 to 102 yielded 48,642 microbubble spatiotemporal trajectory pixels P(x). r , z r , t r 48642 instantaneous blood flow functional feature vectors at (r = 1, 2, 3, ..., 48642) {(D r V r The sequence )|r=1, 2, 3, ..., 48642} was represented in a high-dimensional feature manifold space, forming 48642 instantaneous blood flow function feature points, such as Figure 5A As shown, in the high-dimensional feature manifold space 510 composed of the microbubble instantaneous velocity axis 521, the microbubble instantaneous axial motion direction axis 522, and the microbubble instantaneous transverse motion direction axis 523, pattern recognition is performed on 48642 instantaneous blood flow functional feature points 514, resulting in the following... Figure 5B The blood flow function pattern recognition results shown are as follows, Figure 5A The 48,642 instantaneous blood flow function feature points 514 shown are divided into four blood flow function modes: the first blood flow function mode 525, the second blood flow function mode 526, the third blood flow function mode 527, and the fourth blood flow function mode 528, which contain 11,017, 7,453, 10,219, and 19,953 instantaneous blood flow function feature points, respectively. Each instantaneous blood flow function feature vector belongs to a unique blood flow function mode, that is, each microbubble spatiotemporal trajectory pixel belongs to a unique blood flow function mode.

[0070] In one embodiment, preferably, unsupervised K-means clustering is used for pattern recognition. K-means clustering has a fast convergence speed and high algorithm efficiency, and can adaptively identify blood flow functional patterns with similar functional characteristics.

[0071] In one embodiment, preferably, pattern recognition employs a support vector machine (SVM). SVMs can perform binary classification of high-dimensional data in a non-linear manner, accurately separating blood flow functional patterns with different functional characteristics.

[0072] Step 104: Backtrack the instantaneous blood flow functional features corresponding to the R microbubble spatiotemporal trajectory pixels in the high-dimensional feature manifold space Ω2, which are divided into K classes, to the R microbubble spatiotemporal trajectory pixels in the video space Ω1, and assign a unique color code C to each microbubble spatiotemporal trajectory pixel according to the class. i (i = 1, 2, 3, ..., K); subsequently, K rainbow ultrasound contrast imaging videos Z0 of the same size as the original video Z0 are formed. i (i = 1, 2, 3, ..., K);

[0073] Specifically, Figure 5B The 48,642 microbubble spatiotemporal trajectory pixels in the high-dimensional feature manifold space 520 shown are traced back one by one to, as... Figure 6 The video space 60 shown, and based on the blood flow function mode uniquely assigned to the instantaneous blood flow function feature vectors of 48642 microbubble spatiotemporal trajectory pixels in the high-dimensional feature manifold space 520, are assigned unique color codes to the 48642 microbubble spatiotemporal trajectory pixels in the video space 60, forming four subsets of microbubble spatiotemporal trajectory pixels, such as... Figure 6 As shown, the 11017 microbubble spatiotemporal trajectory pixels in the first blood flow function mode 525 correspond to microbubble spatiotemporal trajectory 64; the 7453 pixels in the second blood flow function mode 526 correspond to microbubble spatiotemporal trajectory 63; the 10219 pixels in the third blood flow function mode 527 correspond to microbubble spatiotemporal trajectories 66-68; and the 19953 pixels in the fourth blood flow function mode 528 correspond to microbubble spatiotemporal trajectories 61, 62, and 65. Furthermore, according to the color coding order, the 11017 microbubble spatiotemporal trajectory pixels belonging to color code 1, i.e., the first blood flow function mode 525, are separated, as follows: Figure 7A As shown, an ultrasound contrast imaging video Z1 with a size of 100×100×300 and a color code of 1, corresponding to the first blood flow function mode 525, is constructed. In this video, 11017 microbubble spatiotemporal trajectory pixels form a microbubble spatiotemporal trajectory 711. The same operation is then performed on the microbubble spatiotemporal trajectory pixels corresponding to the second, third, and fourth blood flow function modes 526, 527, and 528, respectively. In summary, four ultrasound contrast imaging videos Z1, Z2, Z3, and Z4, each with a size of 100×100×300, are obtained, corresponding to... Figure 7A Microbubble spacetime trajectories 711 Figure 7B Microbubble spacetime trajectories 721 Figure 7C The spatiotemporal trajectories of microbubbles in 731-733 Figure 7D The spatiotemporal trajectories of microbubbles (741-743) show that the spatiotemporal trajectories of microbubbles are effectively separated, which lays the foundation for more precise reconstruction and quantification of microvascular structure and function.

[0074] Step 105: Process K rainbow ultrasound contrast images Z iSuper-resolution reconstruction is performed on each of the (i = 1, 2, 3, ..., K) cells, resulting in K color codes denoted as C. i Super-resolution microvessel subset image V (i = 1, 2, 3, ..., K) i (i = 1, 2, 3, ..., K), further, by analyzing all super-resolution microvascular subset images V i Weighted superposition is performed to obtain a rainbow super-resolution microvascular image with K colors;

[0075] Specifically, the corresponding Figure 7A , 7B Super-resolution reconstruction was performed on four ultrasound contrast images (Z1, Z2, Z3, and Z4) of sizes 100×100×300 from images 7C and 7D, respectively, to obtain corresponding results. Figure 8A , 8B The four super-resolution blood vessels 811, 821, 831 and 841 of 8C and 8D, namely the four super-resolution microvascular subset images V1, V2, V3 and V4, are coded as 1, 2, 3 and 4 respectively.

[0076] In one embodiment, preferably, the super-resolution reconstruction strategy employs ultrasound-guided microscopy and nearest neighbor tracking techniques. These techniques offer fast computation speeds, simple algorithm design, and the ability to reconstruct vascular structures and functions with extremely high spatial resolution.

[0077] In one embodiment, preferably, the super-resolution reconstruction strategy employs ultrasound diffraction attenuation microscopy. Ultrasound diffraction attenuation microscopy is an algorithm with high efficiency and fast computation speed, capable of reconstructing continuous vascular structures with extremely high temporal and spatial resolution.

[0078] Step 106: Based on the number of blood flow function patterns K in the high-dimensional feature manifold space Ω2 and the spatial distribution characteristics of the K blood flow function patterns, establish quantitative evaluation indicators for blood flow function, including: the maximum / minimum distance between the centers of the K clusters; and / or the uniformity of the proportion of the K clusters in the total.

[0079] Specifically, in such Figure 5BIn the high-dimensional feature manifold space 520 shown, the average values ​​of 11017, 7453, 10219, and 19953 instantaneous blood flow function feature points corresponding to the first blood flow function mode 525, the second blood flow function mode 526, the third blood flow function mode 527, and the fourth blood flow function mode 528 can be calculated to obtain the center point positions of the four clusters (X1, Y1, Z1), (X2, Y2, Z2), (X3, Y3, Z3), and (X4, Y4, Z4). The proportion of instantaneous blood flow function feature points belonging to the first blood flow function mode 525, the second blood flow function mode 526, the third blood flow function mode 527, and the fourth blood flow function mode 528 to all instantaneous blood flow function feature points can be calculated respectively.

[0080] In one embodiment, the maximum distance between the center points of the four clusters (X1, Y1, Z1), (X2, Y2, Z2), (X3, Y3, Z3), and (X4, Y4, Z4) can be calculated. Specifically, in Figure 5B In the first blood flow functional mode 525, the second blood flow functional mode 526, the third blood flow functional mode 527, and the fourth blood flow functional mode 528 have four cluster center point positions respectively (0.25, 0.43, 0.36), (0.33, 0.17, 0.29), (0.81, -0.27, 0.29), and (0.74, -0.19, 0.52). The Euclidean distances between any two cluster center points are 0.28, 0.90, 0.81, 0.65, 0.59, and 0.25, respectively, with a maximum distance of 0.90.

[0081] In one embodiment, the variance of the proportion of each of the four clusters can be calculated. Specifically, in Figure 5B In the four clusters, the proportions of the instantaneous blood flow function feature points to all instantaneous blood flow function feature points are 0.2, 0.1, 0.6, and 0.1, respectively, and the variance is 0.24.

[0082] Example 2

[0083] Figure 9 The diagram shown is a flowchart illustrating a high-definition vascular structure reconstruction and blood flow function evaluation method for target identification and ultrasound-focused drug release, provided in an embodiment of this application. Figure 9 As shown, this high-definition vascular structure reconstruction and blood flow function evaluation target area identification and ultrasound-focused drug release method includes the following steps:

[0084] Step 901: After preparing drug-loaded microbubbles and injecting them in vivo, a sequence of ultrasound contrast images is acquired using a low mechanical index plane wave emitted by an ultrasound array transducer.

[0085] Specifically, based on the type of disease being treated, a therapeutic drug is selected, and microbubbles loaded with the therapeutic drug are prepared. Further, the drug-carrying ultrasound contrast microbubbles are injected into the organism according to a set drug dosage. Based on the estimated imaging area depth, an ultrasound array transducer with the corresponding imaging center frequency is selected, and the position of the ultrasound probe is fixed at the selected imaging area. The ultrasound array transducer is then activated to emit a plane wave with a low mechanical index to acquire plane wave ultrasound data and obtain an M-frame of ultrasound contrast imaging images.

[0086] It should be noted that drug-loaded microbubbles refer to microbubbles capable of carrying drugs, typically with a diameter of less than 5 micrometers. The drug carried can be any compound with therapeutic or preventative effects, including gene therapy drugs, chemotherapy drugs, or targeted antibody drugs. The drug can be loaded onto the surface of the microbubble or encapsulated within it; this application does not impose specific limitations. Furthermore, drug-loaded microbubbles can generate specific acoustic responses under ultrasound stimulation, such as generating harmonics and subharmonics, which can then be visualized in ultrasound images. The injection method for drug-loaded microbubbles can be intravenous injection, or injection into lymphatic vessels or ureters; this application does not impose specific limitations.

[0087] In one embodiment of this application, the array transducer is an array of multiple transducers arranged in a certain form, such as a linear array or a phased array, to complete the transmission, focusing, and reception of ultrasonic waves. During the acquisition of ultrasound contrast imaging images, dynamic video images of microbubble flow can be displayed in real time on an ultrasound imaging monitor.

[0088] In one embodiment of this application, the mechanical index is set to 0.1 and the imaging frame rate is set to 500Hz.

[0089] In one embodiment of this application, a high-frequency probe with an imaging center frequency of 15MHz is selected for shallower imaging depths, such as lymph nodes and thyroid glands; and a low-frequency probe with an imaging center frequency of 7MHz is selected for deeper imaging depths, such as kidneys and livers.

[0090] Step 902: Identify the drug release target area.

[0091] Specifically, by acquiring M-frame contrast-enhanced ultrasound images after drug-loaded microbubbles are released, the blood flow structure and functional characteristics of the imaging area are analyzed to identify the range of drug-targeted vascular regions. The target vascular region for drug release can be selected manually based on high-resolution vascular reconstruction results, or automatically based on lesion characteristics. This application does not specify the number of ultrasound contrast-enhanced image sequences or target vascular regions to be acquired.

[0092] The analysis of blood flow structure and functional characteristics in the imaging region may include the following steps: The original video Z0, containing M frames of ultrasound contrast images, is stitched together chronologically to obtain a video space Ω1. Further, R microbubble spatiotemporal trajectory pixels are obtained by identifying the pixels to which microbubbles belong. Within the video space Ω1, the instantaneous velocity and direction of the microbubble at each microbubble spatiotemporal trajectory pixel are estimated, i.e., the instantaneous blood flow functional characteristics. Then, a high-dimensional feature manifold space Ω2 is constructed based on the instantaneous blood flow functional characteristics corresponding to the R microbubble spatiotemporal trajectory pixels. Within the 2D feature manifold space Ω2, pattern recognition methods are used to classify the instantaneous blood flow functional features corresponding to R microbubble spatiotemporal trajectory pixels into K classes, i.e., K types of blood flow functional modes. Each microbubble spatiotemporal trajectory pixel uniquely belongs to a particular blood flow functional mode. The instantaneous blood flow functional features corresponding to the R microbubble spatiotemporal trajectory pixels classified into K classes in the 2D feature manifold space Ω2 are then backtracked to the R microbubble spatiotemporal trajectory pixels in the video space Ω1, and each microbubble spatiotemporal trajectory pixel is assigned a unique color code C based on its category. i (i = 1, 2, 3, ..., K); subsequently, K rainbow ultrasound contrast imaging videos Z0 are formed, each with the same size as the original video Z0 and reflecting different blood flow functional characteristics. i (i = 1, 2, 3, ..., K). The number of categories K and the inter-class differences are used as indicators of vascular heterogeneity within the scanning field of view.

[0093] Specifically, the manual interactive selection based on high-definition vascular reconstruction results includes the following processing steps: Z-banding of rainbow ultrasound contrast-enhanced videos with different blood flow functional characteristics. i (i = 1, 2, 3, ..., K) are reconstructed using high-resolution microvessels to form rainbow super-resolution microvessel images with K colors, i.e., high-resolution vascular structure videos Z. * The location and size of the target blood vessel area for drug release are displayed on the screen, allowing users to judge and interactively select the location.

[0094] The high-definition microvascular reconstruction strategy preferably employs ultrasound-guided microscopy. Ultrasound-guided microscopy boasts fast computation speed and simple algorithms, enabling the reconstruction of vascular structure and function with extremely high spatial resolution; however, this application does not impose specific limitations on it.

[0095] It should be noted that interactive area selection can be done via touchpad or mouse input, and the shape of the selected area can be marked with a square or circle. The number of selected areas can be one or more, and this application does not impose any specific limitations.

[0096] In one embodiment of this application, the user can manually select an interactive region based on a single super-resolution microvascular image, or select from K color-coded C iSuper-resolution microvessel subset image V (i = 1, 2, 3, ..., K) i (i = 1, 2, 3, ..., K) Select a portion of the image for weighted overlay, and then perform manual interactive region selection based on the overlaid image.

[0097] In another embodiment of this application, the user can first manually and interactively select multiple vascular regions in multiple super-resolution microvascular images. Then, the computer performs quantitative analysis of the vascular structure and functional parameters of the selected vascular regions and provides quantitative values ​​back to the user. Finally, the user filters vascular regions based on the feedback values ​​to determine the target vascular region.

[0098] It should be noted that the vascular structure and function parameters can be regional vascular density, regional average vascular tortuosity, regional vascular tortuosity distribution, regional average vascular diameter, regional vascular diameter distribution, regional average blood flow velocity, regional blood flow velocity distribution, and regional blood flow direction entropy. This application does not impose specific limitations.

[0099] Specifically, the automatic selection process based on lesion characteristics includes the following steps: automatically selecting the two-dimensional imaging section with the highest heterogeneity index, and automatically extracting the rainbow ultrasound contrast video Z with the highest microbubble flow rate and corresponding color code c. c (c < K), and perform maximum density projection operation on it, selecting the set of coordinates of pixels with gray values ​​greater than 0 after projection as the target blood vessel region. In addition, automatically extract the P rainbow ultrasound contrast images Z1, Z2, ..., Zn based on the top P microbubble flow rates. P (P < K), and perform maximum density projection operation on them respectively, and then process the P rainbow projection images MIP1, MIP2, ..., MIP after the maximum projection. P (P < K) Perform a second maximum projection operation, and select the set of coordinates of pixels with gray values ​​greater than 0 after the second maximum projection as the target blood vessel region. This application does not impose specific limitations.

[0100] Step 903: Activate the ultrasonic array transducer to emit ultrasonic waves with a high mechanical index, focused on the target release area, to break up the drug-loaded microbubbles.

[0101] The focus control methods include single focus, multifocus and variable focus. Specifically, the size, position and number of basic focus units to be synthesized are determined according to the size and location of the lesion area. Considering that the target vascular area is a certain range, this application chooses the multifocus form to achieve the destruction of drug-loaded microbubbles in the target vascular area.

[0102] It should be noted that the device emitting focused ultrasound waves can be an ultrasound array transducer used for ultrasound contrast imaging, or a second ultrasound array transducer specifically for focused imaging; this application does not impose any specific limitations. Users can change the focal spot size by altering the number of array elements involved in the focusing process. By selecting an ultrasound array transducer with an appropriate number of array elements, drug release can be achieved within a larger or smaller local area by switching between different focal spot sizes. Users can select an appropriate focal spot size based on the shape and size of the target region; this application does not impose any specific limitations.

[0103] In one embodiment of this application, the sound pressure of the focused ultrasound is set to 0.3 MPa, the number of cycles is 5000, the ultrasound treatment time is 4 minutes, and the mechanical index is 0.3.

[0104] Step 904: Initiate plane wave acquisition under low mechanical index to obtain ultrasound contrast images of the target area after drug treatment, and evaluate the effect of the drug.

[0105] After the drug release operation is completed, the position of the ultrasound probe is fixed at the target release area, and the ultrasound array transducer is activated to emit a plane wave with a low mechanical index. Plane wave ultrasound data is acquired to obtain M frames of ultrasound contrast images, and the drug effect is quantitatively evaluated based on the ultrasound contrast images.

[0106] Specifically, quantitative evaluation of drug efficacy includes: for M-frame ultrasound contrast images of the target release area, taking the mean or median gray value of the image respectively, to obtain the echo intensity change function over time, i.e., the time-intensity curve. By fitting the time-intensity curve, vascular function evaluation parameters can be obtained, such as peak time and area under the curve.

[0107] In another embodiment of this application, the quantitative evaluation of the drug effect includes: stitching together the original video Z0 containing M frames of ultrasound contrast images in chronological order to obtain a video space Ω1; further, identifying the pixels to which microbubbles belong to obtain a total of R microbubble spatiotemporal trajectory pixels; within the video space Ω1, estimating the instantaneous velocity and direction of microbubble motion at each microbubble spatiotemporal trajectory pixel, i.e., instantaneous blood flow function characteristics; then, constructing a high-dimensional feature manifold space Ω2 based on the instantaneous blood flow function characteristics corresponding to the R microbubble spatiotemporal trajectory pixels; within the high-dimensional feature manifold space Ω2, using a pattern recognition method to classify the instantaneous blood flow function characteristics corresponding to the R microbubble spatiotemporal trajectory pixels into K categories, i.e., K types of blood flow function patterns; further, based on the number K of blood flow function patterns K within the high-dimensional feature manifold space Ω2 and the spatial distribution of the K blood flow function patterns, calculating the maximum / minimum distance between the centers of the K clusters; and / or the uniformity of the overall proportion of the K clusters, as parameters for evaluating the drug effect.

[0108] Example 3

[0109] Figure 10 The diagram shown is a schematic representation of a high-definition vascular structure reconstruction and blood flow function evaluation device provided in an embodiment of this application. Figure 10 As shown, the high-definition vascular structure reconstruction and blood flow function evaluation device 1000 includes:

[0110] The video space construction module 1010 is configured to stitch together the original video Z0 containing M frames of ultrasound contrast images in chronological order to obtain the video space Ω1. Furthermore, a total of R microbubble spatiotemporal trajectory pixels are obtained by identifying the pixels to which the microbubbles belong.

[0111] The high-dimensional feature manifold space construction module 1020 estimates the instantaneous motion speed and direction of each microbubble spatiotemporal trajectory pixel in the video space Ω1, i.e. the instantaneous blood flow function features. Then, based on the instantaneous blood flow function features corresponding to R microbubble spatiotemporal trajectory pixels, the high-dimensional feature manifold space Ω2 is constructed.

[0112] The blood flow function pattern recognition module 1030 uses a pattern recognition method to classify the instantaneous blood flow function features corresponding to R microbubble spatiotemporal trajectory pixels into K classes, i.e. K kinds of blood flow function patterns, in the high-dimensional feature manifold space Ω2. Among them, the blood flow function feature corresponding to each microbubble spatiotemporal trajectory pixel belongs to a unique blood flow function pattern.

[0113] The Rainbow Ultrasound Contrast Imaging Video Reconstruction Module 1040 backtracks the instantaneous blood flow functional features corresponding to the R microbubble spatiotemporal trajectory pixels in the high-dimensional feature manifold space Ω2, which are divided into K classes, to the R microbubble spatiotemporal trajectory pixels in the video space Ω1, and assigns a unique color code C to each microbubble spatiotemporal trajectory pixel according to its class. i (i = 1, 2, 3, ..., K); subsequently, K rainbow ultrasound contrast imaging videos Z0 of the same size as the original video Z0 are formed. i (i = 1, 2, 3, ..., K);

[0114] High-definition vascular structure video reconstruction module 1050: This module reconstructs rainbow ultrasound contrast-enhanced video. i High-resolution vascular structure video Z was obtained using a super-resolution reconstruction algorithm. * ;

[0115] Vascular heterogeneity assessment module 1060: The number of categories K within the scanning field of view and the inter-class differences are used as indicators of vascular heterogeneity.

[0116] In one embodiment, the video space construction module 1010 may be further configured to: stitch together the original video Z0 (image size A×L, unit: pixels) containing M frames of two-dimensional ultrasound contrast images frame by frame in the form of a numerical matrix and in chronological order to construct a video space Ω1 (three-dimensional, A×L×M, unit: pixels); and identify R microbubble spatiotemporal trajectory pixels P(x) within the video space Ω1 whose grayscale values ​​are greater than a set grayscale threshold. r , z r , t r )(r=1,2,3,...,R,1≤x r ≤A, 1≤y r ≤L,1≤z r ≤M) are identified as pixels belonging to the spatiotemporal trajectory of microbubbles, wherein the R microbubble spatiotemporal trajectory pixels P(x r , z r , t r It can present the three-dimensional spatiotemporal trajectory of moving microbubbles in the video space Ω1.

[0117] In one embodiment, the high-dimensional feature manifold space construction module 1020 may be further configured to: in the video space Ω1, based on R microbubble spatiotemporal trajectory pixels P(x r , z r , t r The distribution gradient of local gray values ​​at (r = 1, 2, 3, ..., R) is calculated, and the local structure tensor is calculated; and / or the trajectory skeleton is extracted and a straight line is fitted to obtain R microbubble spatiotemporal trajectory pixels P(x) r , z r , t r The instantaneous trajectory direction vector T at (r = 1, 2, 3, ..., R) r (x r , z r , t r (r = 1, 2, 3, ..., R); Calculate the trajectory direction vector T for each instantaneous moment. r (x r , z r , t r The azimuth angle θ corresponding to (r = 1, 2, 3, ..., R) r (x r , z r , t r (r = 1, 2, 3, ..., R) and polar angle The spatiotemporal trajectories of R microbubbles in the video space Ω1 are obtained as follows: P(x) r , z r , t r The instantaneous motion direction D of the microbubbles at (r = 1, 2, 3, ..., R)r (r = 1, 2, 3, ..., R) and instantaneous velocity V r (r = 1, 2, 3, ..., R); P(x) represents the R microbubble spatiotemporal trajectory pixels in the video space Ω1. r , z r , t r The instantaneous motion direction D of the microbubbles at (r = 1, 2, 3, ..., R) r (r = 1, 2, 3, ..., R) and instantaneous velocity V r (r = 1, 2, 3, ..., R) are paired to form the instantaneous blood flow functional feature vector {(D r V r )|r=1,2,3,...,R} and mapped to the high-dimensional feature manifold space Ω2.

[0118] In one embodiment, the blood flow function pattern recognition module 1030 may be further configured to: within the high-dimensional feature manifold space Ω2, for the instantaneous blood flow function feature vectors {(D... r V r Unsupervised or weakly supervised clustering is performed on the sequence |r=1,2,3,...,R} to obtain K clusters, which are K types of blood flow function patterns. Each instantaneous blood flow function feature vector belongs to a unique blood flow function pattern, that is, each microbubble spatiotemporal trajectory pixel belongs to a unique blood flow function pattern.

[0119] In one embodiment, the rainbow ultrasound contrast imaging video reconstruction module 1040 can be further configured to: trace back R microbubble spatiotemporal trajectory pixels from the high-dimensional feature manifold space Ω2 to the video space Ω1 one by one, and based on the R microbubble spatiotemporal trajectory pixels P(x) in the high-dimensional feature manifold space Ω2... r , z r , t r The instantaneous blood flow functional feature vector {(D)} r V r The unique blood flow function mode of the sequence Ω1 (r = 1, 2, 3, ..., R) assigns a unique color code C to the corresponding microbubble spatiotemporal trajectory pixel in the video space Ω1. i (i = 1, 2, 3, ..., K), ultimately forming K subsets of microbubble spatiotemporal trajectory pixels; according to the color encoding order, the pixels in the original video Z0 belonging to color encoding C... i The subset of pixels representing the spatiotemporal trajectory of the i-th microbubble (i = 1, 2, 3, ..., K) is sequentially separated to construct K rainbow ultrasound contrast imaging videos Z0 of the same size as the original video Z0. i (i = 1, 2, 3, ..., K).

[0120] In one embodiment, the high-definition vascular structure video reconstruction module 1050 may be further configured to: reconstruct K rainbow ultrasound contrast images Z i (i = 1, 2, 3, ..., K) are reconstructed using ultrasound localization microscopy, resulting in K color codes coded as C. i Super-resolution microvessel subset image V (i = 1, 2, 3, ..., K) i (i = 1, 2, 3, ..., K), further, by analyzing all super-resolution microvascular subset images V i Weighted superposition yields a rainbow-like super-resolution microvascular image with K colors, i.e., a high-definition vascular structure video Z. * .

[0121] In one embodiment, the vascular heterogeneity evaluation module 1060 may be further configured to: calculate the maximum / minimum distance between the centers of the K clusters based on the number K blood flow functional patterns K in the high-dimensional feature manifold space Ω2 and the spatial distribution of the K blood flow functional patterns; and / or the uniformity of the proportion of the K clusters in the total.

[0122] It should be noted that the high-definition vascular structure reconstruction and blood flow function evaluation device 1000 according to the embodiments of this application is integrated into the electronic device 1100 as a software module and / or hardware module. In other words, the electronic device 1100 includes the high-definition vascular structure reconstruction and blood flow function evaluation device 1000. For example, the high-definition vascular structure reconstruction and blood flow function evaluation device 1000 is a software module in the operating system of the electronic device 1100, or an application developed for it.

[0123] In another embodiment of this application, the high-definition vascular structure reconstruction and blood flow function evaluation device 1000 and the electronic device 1100 may also be separate devices (e.g., a server), and the high-definition vascular structure reconstruction and blood flow function evaluation device 1000 may be connected to the electronic device 1100 via wired and / or wireless networks, and transmit interactive information in accordance with an agreed data format.

[0124] Example 4

[0125] Figure 11 The diagram shown is a structural schematic of an electronic device provided in an embodiment of this application. Figure 11 As shown, the electronic device 1100 includes: one or more processors 1101 and a memory 1102; and computer program instructions stored in the memory 1102, which, when executed by the processor 1101, cause the processor 1101 to perform the high-definition vascular structure reconstruction and blood flow function evaluation method as described in any of the above embodiments.

[0126] The processor 1101 may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions.

[0127] The memory 1102 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 1101 may execute the program instructions to implement the steps and / or other desired functions in the high-definition vascular structure reconstruction and blood flow function evaluation methods of the various embodiments of this application described above. Information such as microbubble angiography images and microbubble velocities may also be stored in the computer-readable storage medium.

[0128] In one example, electronic device 1100 includes an input device 1103 and an output device 1104, which are interconnected via a bus system and / or other forms of connection mechanism.

[0129] For example, when the electronic device is an ultrasound imaging system, the input device 1103 can be an ultrasound probe used to detect the position of microbubbles. When the electronic device is a standalone device, the input device 1103 can be a communication network connector used to receive the acquired input signals from an external mobile device. Furthermore, the input device 1103 may also include, for example, a keyboard, mouse, microphone, etc.

[0130] The output device 1104 can output various information to the outside, such as a display, speaker, printer, communication network and its connected remote output devices, etc.

[0131] Of course, for the sake of simplicity, Figure 11 Only some of the components of the electronic device 1100 relevant to this application are shown in this illustration, omitting components such as buses, input devices / output interfaces, etc. In addition, the electronic device 1100 may include any other suitable components depending on the specific application.

[0132] Example 5

[0133] In addition to the methods and devices described above, embodiments of this application may also be computer program products, including computer program instructions, which, when executed by a processor, cause the processor to perform the steps in the high-definition vascular structure reconstruction and blood flow function evaluation method as described in any of the above embodiments.

[0134] Computer program products can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this application. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, and can also include conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0135] Furthermore, embodiments of this application may also be computer-readable storage media storing computer program instructions that, when executed by a processor, cause the processor to perform the steps in the high-definition vascular structure reconstruction and blood flow function evaluation methods according to various embodiments of this application as described in the "High-Definition Vascular Structure Reconstruction and Blood Flow Function Evaluation Method" section of this specification.

[0136] Computer-readable storage media may take the form of any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may, for example, include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0137] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0138] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used in this application refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used in this application refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0139] It should also be noted that in the apparatus, equipment, and methods of this application, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of this application.

[0140] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0141] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

[0142] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications or equivalent substitutions made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for high-definition vascular structure reconstruction and blood flow function evaluation, characterized in that, Based on ultrasound contrast imaging, a high-dimensional feature manifold space is constructed to identify vascular subgroups with different blood flow functional patterns, ultimately obtaining high-resolution rainbow vascular structure videos and vascular heterogeneity indicators within the ultrasound scanning area; the specific steps include: (1) The original video Z0 containing M frames of ultrasound contrast images is spliced ​​together in time order to obtain the video space Ω1, and a total of R microbubble spatiotemporal trajectory pixels are obtained by identifying the pixels to which the microbubbles belong. (2) In the video space Ω1, estimate the instantaneous motion speed and direction of each microbubble at the spatiotemporal trajectory pixel point, i.e. the instantaneous blood flow function features. Then, based on the instantaneous blood flow function features corresponding to R microbubble spatiotemporal trajectory pixels, construct the high-dimensional feature manifold space Ω2. (3) In the high-dimensional feature manifold space Ω2, the instantaneous blood flow function features corresponding to R microbubble spatiotemporal trajectory pixels are divided into K classes, i.e. K blood flow function modes, by using the pattern recognition method. Among them, the blood flow function mode to which the instantaneous blood flow function feature corresponding to each microbubble spatiotemporal trajectory pixel belongs is unique. (4) The instantaneous blood flow function features corresponding to the R microbubble spatiotemporal trajectory pixels in the high-dimensional feature manifold space Ω2, which are divided into K classes, are backtracked to the R microbubble spatiotemporal trajectory pixels in the video space Ω1, and each microbubble spatiotemporal trajectory pixel is assigned a unique color code C according to the category. i Furthermore, K rainbow ultrasound contrast imaging videos Z0, each the same size as the original video Z0 and reflecting different blood flow functional characteristics, are generated. i ; (5) Rainbow ultrasound contrast video Z i Super-resolution reconstruction was performed to obtain a high-definition video of the vascular structure. * ; (6) The number of categories K and the inter-class differences are used as indicators of vascular heterogeneity in the scanning field of view.

2. The method according to claim 1, wherein the construction of the video space Ω1 comprises: (1) For the original video Z0 containing M frames of two-dimensional ultrasound contrast images, the two-dimensional ultrasound contrast images are stitched together frame by frame in the form of a numerical matrix and in chronological order to construct the video space Ω1. (2) R microbubble spatiotemporal trajectory pixels P with gray values ​​greater than a set gray threshold in the video space Ω1 are identified as pixels belonging to the microbubble spatiotemporal trajectory, wherein the R microbubble spatiotemporal trajectory pixels P can present the three-dimensional spatiotemporal trajectory of moving microbubbles in the video space Ω1.

3. The method according to claim 1, wherein the construction of the high-dimensional feature manifold space Ω2 comprises: Based on the instantaneous motion direction and instantaneous motion rate, the R microbubble spatiotemporal trajectory pixels in the video space Ω1 are mapped and represented in the high-dimensional feature manifold space Ω2 through the following steps: (1) In the video space Ω1, calculate the local structure tensor based on the distribution gradient of the local gray values ​​at the R microbubble spatiotemporal trajectory pixel point P; and / or extract the trajectory skeleton and perform straight line fitting to obtain the instantaneous trajectory direction vector T at the R microbubble spatiotemporal trajectory pixel point P. r ; (2) Calculate the trajectory direction vector T for each instant. r The corresponding azimuth angle θ r and polar angle The instantaneous motion direction D of the microbubble at pixel P, representing the spatiotemporal trajectory of R microbubbles in video space Ω1, is obtained. r and instantaneous velocity V r ; (3) The instantaneous motion direction D of the microbubble at pixel point P, which is one of the R microbubble spatiotemporal trajectories in the video space Ω1. r and instantaneous velocity V r Pairs are combined to form R instantaneous blood flow functional feature vectors {(D r V r )} and mapped to the high-dimensional feature manifold space Ω2.

4. According to claim 1, the K blood flow function modes are obtained by the following steps: within the high-dimensional feature manifold space Ω2, for the R instantaneous blood flow function feature vectors {(D... r V r Unsupervised or weakly supervised clustering is performed to obtain K clusters, which represent K blood flow function patterns. Each instantaneous blood flow function feature vector belongs to a unique blood flow function pattern, that is, each microbubble spatiotemporal trajectory pixel belongs to a unique blood flow function pattern.

5. The method according to claim 1, wherein the rainbow ultrasound contrast video is obtained by the following steps: (1) The R microbubble spatiotemporal trajectory pixels are traced back from the high-dimensional feature manifold space Ω2 to the video space Ω1 one by one, and the instantaneous blood flow function feature vector {(D) of the R microbubble spatiotemporal trajectory pixels P in the high-dimensional feature manifold space Ω2 is used as the basis for the calculation. r V r The unique blood flow function mode assigns a unique color code C to the corresponding microbubble spatiotemporal trajectory pixel in video space Ω1. i Ultimately, this forms a subset of K microbubble spatiotemporal trajectory pixels; (2) According to the color coding order, the original video Z0 belongs to color coding C i The subset of pixels representing the spatiotemporal trajectory of the i-th microbubble is sequentially separated to construct K rainbow ultrasound contrast imaging videos Z0, each of the same size as the original video Z0 and reflecting different blood flow functional characteristics. i .

6. A high-definition vascular structure reconstruction and blood flow function evaluation system, characterized in that, The system using the method described in claim 1 comprises the following modules: (1) Video space construction module: The original video Z0 containing M frames of ultrasound contrast images is spliced ​​together in time order to obtain video space Ω1, and a total of R microbubble spatiotemporal trajectory pixels are obtained by identifying the pixels to which the microbubbles belong. (2) High-dimensional feature manifold space construction module: In the video space Ω1, estimate the instantaneous motion speed and direction of each microbubble spatiotemporal trajectory pixel point, i.e. instantaneous blood flow function features. Then, based on the instantaneous blood flow function features corresponding to R microbubble spatiotemporal trajectory pixels points, construct the high-dimensional feature manifold space Ω2. (3) Blood flow function pattern recognition module: In the high-dimensional feature manifold space Ω2, the instantaneous blood flow function features corresponding to R microbubble spatiotemporal trajectory pixels are divided into K classes, i.e. K kinds of blood flow function patterns, in which the blood flow function feature corresponding to each microbubble spatiotemporal trajectory pixel belongs to a unique blood flow function pattern. (4) The Rainbow Ultrasound Contrast Imaging Video Reconstruction Module backtracks the instantaneous blood flow functional features corresponding to the R microbubble spatiotemporal trajectory pixels in the high-dimensional feature manifold space Ω2, which are divided into K categories, to the R microbubble spatiotemporal trajectory pixels in the video space Ω1, and assigns a unique color code C to each microbubble spatiotemporal trajectory pixel according to the category. i Furthermore, K rainbow ultrasound contrast imaging videos Z0, each the same size as the original video Z0 and reflecting different blood flow functional characteristics, are generated. i ; (5) High-definition vascular structure video reconstruction module: This module reconstructs the rainbow ultrasound contrast imaging video. i High-resolution vascular structure video Z was obtained using a super-resolution reconstruction algorithm. * ; (6) Vascular heterogeneity assessment module: The number of categories K in the scanning field of view and the inter-class differences are used as indicators of vascular heterogeneity.

7. An electronic device, comprising: processor; as well as A memory storing computer program instructions that, when executed by the processor, cause the processor to perform the high-definition vascular structure reconstruction and blood flow function evaluation method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, comprising: A computer-readable storage medium storing computer program instructions that, when executed by a processor, cause the processor to perform the high-definition vascular structure reconstruction and blood flow function evaluation method as described in any one of claims 1 to 5.

Images